Robust Watermarking Techniques against Compression Attack

Abstract

Digital watermarking is a way to protect digital images from malicious attacks. Compression attack is one of the most common attacks on images uploaded to social media. Social media, such as Facebook and Twitter, have implemented a compression method for all types of media before being successfully uploaded to their server. It is designed to reduce the network bandwidth and storage needed to store each media on their server. However, the compression method used tends to tarnish the image properties of the image itself, which can be used to identify the image itself. This causes another problem, namely ownership and copyright issues. The use of digital watermarks was therefore proposed for this research to avoid the problem identified. The selected digital watermarking techniques, which are DCT-based Pyramid Transform (DCTPT), Two-Step Sudoku Method using LSB (TSSLSB), Semi-fragile Spatial Watermarking based on Local Binary Pattern Operator (SSWLBP) and Hybrid Schur and Singular Value Decomposition (HSSVD), theoretically must be able to withstand compression attacks by social media. These techniques have been simulated using the default MATLAB image dataset, but the use of photos taken by smartphones is a must for us to obtain useful metadata for real-world implementations. Some comprehensive analyses of the watermarking algorithms were done, and the findings were recorded in this paper.